Jacky Zhao
2024-07-10 247625c4f538a113617e553e1ed3c74d94545e62
quartz/plugins/transformers/ofm.ts
@@ -1,11 +1,11 @@
import { QuartzTransformerPlugin } from "../types"
import { Blockquote, Root, Html, BlockContent, DefinitionContent, Paragraph, Code } from "mdast"
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 { SKIP, visit } from "unist-util-visit"
import path from "path"
import { splitAnchor } from "../../util/path"
import { JSResource } from "../../util/resources"
// @ts-ignore
import calloutScript from "../../components/scripts/callout.inline.ts"
@@ -17,7 +17,6 @@
import { PhrasingContent } from "mdast-util-find-and-replace/lib"
import { capitalize } from "../../util/lang"
import { PluggableList } from "unified"
import { ValidCallout, i18n } from "../../i18n"
export interface Options {
  comments: boolean
@@ -98,32 +97,40 @@
export const externalLinkRegex = /^https?:\/\//i
export const arrowRegex = new RegExp(/(-{1,2}>|={1,2}>|<-{1,2}|<={1,2})/, "g")
export const arrowRegex = new RegExp(/(-{1,2}>|={1,2}>|<-{1,2}|<={1,2})/g)
// !?                -> 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)
// !?                 -> optional embedding
// \[\[               -> open brace
// ([^\[\]\|\#]+)     -> one or more non-special characters ([,],|, or #) (name)
// (#[^\[\]\|\#]+)?   -> # then one or more non-special characters (heading link)
// (\\?\|[^\[\]\#]+)? -> optional escape \ then | then one or more non-special characters (alias)
export const wikilinkRegex = new RegExp(
  /!?\[\[([^\[\]\|\#]+)?(#+[^\[\]\|\#]+)?(\|[^\[\]\#]+)?\]\]/,
  "g",
  /!?\[\[([^\[\]\|\#\\]+)?(#+[^\[\]\|\#\\]+)?(\\?\|[^\[\]\#]+)?\]\]/g,
)
const highlightRegex = new RegExp(/==([^=]+)==/, "g")
const commentRegex = new RegExp(/%%[\s\S]*?%%/, "g")
// ^\|([^\n])+\|\n(\|) -> matches the header row
// ( ?:?-{3,}:? ?\|)+  -> matches the header row separator
// (\|([^\n])+\|\n)+   -> matches the body rows
export const tableRegex = new RegExp(/^\|([^\n])+\|\n(\|)( ?:?-{3,}:? ?\|)+\n(\|([^\n])+\|\n?)+/gm)
// matches any wikilink, only used for escaping wikilinks inside tables
export const tableWikilinkRegex = new RegExp(/(!?\[\[[^\]]*?\]\])/g)
const highlightRegex = 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")
const calloutRegex = new RegExp(/^\[\!(\w+)\|?(.+?)?\]([+-]?)/)
const calloutLineRegex = new RegExp(/^> *\[\!\w+\|?.*?\][+-]?.*$/gm)
// (?:^| )              -> non-capturing group, tag should start be separated by a space or be the start of the line
// #(...)               -> capturing group, tag itself must start with #
// (?:[-_\p{L}\d\p{Z}])+       -> non-capturing group, non-empty string of (Unicode-aware) alpha-numeric characters and symbols, hyphens and/or underscores
// (?:\/[-_\p{L}\d\p{Z}]+)*)   -> non-capturing group, matches an arbitrary number of tag strings separated by "/"
const tagRegex = new RegExp(
  /(?:^| )#((?:[-_\p{L}\p{Emoji}\p{M}\d])+(?:\/[-_\p{L}\p{Emoji}\p{M}\d]+)*)/,
  "gu",
  /(?:^| )#((?:[-_\p{L}\p{Emoji}\p{M}\d])+(?:\/[-_\p{L}\p{Emoji}\p{M}\d]+)*)/gu,
)
const blockReferenceRegex = new RegExp(/\^([-_A-Za-z0-9]+)$/, "g")
const blockReferenceRegex = new RegExp(/\^([-_A-Za-z0-9]+)$/g)
const ytLinkRegex = /^.*(youtu.be\/|v\/|u\/\w\/|embed\/|watch\?v=|\&v=)([^#\&\?]*).*/
const ytPlaylistLinkRegex = /[?&]list=([^#?&]*)/
const videoExtensionRegex = new RegExp(/\.(mp4|webm|ogg|avi|mov|flv|wmv|mkv|mpg|mpeg|3gp|m4v)$/)
const wikilinkImageEmbedRegex = new RegExp(
  /^(?<alt>(?!^\d*x?\d*$).*?)?(\|?\s*?(?<width>\d+)(x(?<height>\d+))?)?$/,
@@ -169,13 +176,27 @@
          src = src.toString()
        }
        // replace all wikilinks inside a table first
        src = src.replace(tableRegex, (value) => {
          // escape all aliases and headers in wikilinks inside a table
          return value.replace(tableWikilinkRegex, (_value, raw) => {
            // const [raw]: (string | undefined)[] = capture
            let escaped = raw ?? ""
            escaped = escaped.replace("#", "\\#")
            // escape pipe characters if they are not already escaped
            escaped = escaped.replace(/((^|[^\\])(\\\\)*)\|/g, "$1\\|")
            return escaped
          })
        })
        // replace all other wikilinks
        src = src.replace(wikilinkRegex, (value, ...capture) => {
          const [rawFp, rawHeader, rawAlias]: (string | undefined)[] = capture
          const fp = rawFp ?? ""
          const anchor = rawHeader?.trim().replace(/^#+/, "")
          const [fp, anchor] = splitAnchor(`${rawFp ?? ""}${rawHeader ?? ""}`)
          const blockRef = Boolean(anchor?.startsWith("^")) ? "^" : ""
          const displayAnchor = anchor ? `#${blockRef}${slugAnchor(anchor)}` : ""
          const displayAnchor = anchor ? `#${blockRef}${anchor.trim().replace(/^#+/, "")}` : ""
          const displayAlias = rawAlias ?? rawHeader?.replace("#", "|") ?? ""
          const embedDisplay = value.startsWith("!") ? "!" : ""
@@ -189,9 +210,8 @@
      return src
    },
    markdownPlugins(ctx) {
    markdownPlugins(_ctx) {
      const plugins: PluggableList = []
      const cfg = ctx.cfg.configuration
      // regex replacements
      plugins.push(() => {
@@ -250,7 +270,7 @@
                    return {
                      type: "html",
                      data: { hProperties: { transclude: true } },
                      value: `<blockquote class="transclude" data-url="${url}" data-block="${block}"><a href="${
                      value: `<blockquote class="transclude" data-url="${url}" data-block="${block}" data-embed-alias="${alias}"><a href="${
                        url + anchor
                      }" class="transclude-inner">Transclude of ${url}${block}</a></blockquote>`,
                    }
@@ -328,7 +348,7 @@
                  children: [
                    {
                      type: "text",
                      value: `#${tag}`,
                      value: tag,
                    },
                  ],
                }
@@ -388,8 +408,8 @@
                return
              }
              // find first line
              const firstChild = node.children[0]
              // find first line and callout content
              const [firstChild, ...calloutContent] = node.children
              if (firstChild.type !== "paragraph" || firstChild.children[0]?.type !== "text") {
                return
              }
@@ -401,7 +421,7 @@
              const match = firstLine.match(calloutRegex)
              if (match && match.input) {
                const [calloutDirective, typeString, collapseChar] = match
                const [calloutDirective, typeString, calloutMetaData, collapseChar] = match
                const calloutType = canonicalizeCallout(typeString.toLowerCase())
                const collapse = collapseChar === "+" || collapseChar === "-"
                const defaultState = collapseChar === "-" ? "collapsed" : "expanded"
@@ -463,8 +483,24 @@
                    className: classNames.join(" "),
                    "data-callout": calloutType,
                    "data-callout-fold": collapse,
                    "data-callout-metadata": calloutMetaData,
                  },
                }
                // Add callout-content class to callout body if it has one.
                if (calloutContent.length > 0) {
                  const contentData: BlockContent | DefinitionContent = {
                    data: {
                      hProperties: {
                        className: "callout-content",
                      },
                      hName: "div",
                    },
                    type: "blockquote",
                    children: [...calloutContent],
                  }
                  node.children = [node.children[0], contentData]
                }
              }
            })
          }
@@ -528,12 +564,35 @@
                    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,
                    if (last.value === "") {
                      // this is an inline block ref but the actual block
                      // is the previous element above it
                      let idx = (index ?? 1) - 1
                      while (idx >= 0) {
                        const element = parent?.children.at(idx)
                        if (!element) break
                        if (element.type !== "element") {
                          idx -= 1
                        } else {
                          if (!Object.keys(file.data.blocks!).includes(block)) {
                            element.properties = {
                              ...element.properties,
                              id: block,
                            }
                            file.data.blocks![block] = element
                          }
                          return
                        }
                      }
                      file.data.blocks![block] = node
                    } else {
                      // normal paragraph transclude
                      if (!Object.keys(file.data.blocks!).includes(block)) {
                        node.properties = {
                          ...node.properties,
                          id: block,
                        }
                        file.data.blocks![block] = node
                      }
                    }
                  }
                }
@@ -552,7 +611,9 @@
              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
                const playlistId = node.properties.src.match(ytPlaylistLinkRegex)?.[1]
                if (videoId) {
                  // YouTube video (with optional playlist)
                  node.tagName = "iframe"
                  node.properties = {
                    class: "external-embed",
@@ -560,7 +621,20 @@
                    frameborder: 0,
                    width: "600px",
                    height: "350px",
                    src: `https://www.youtube.com/embed/${videoId}`,
                    src: playlistId
                      ? `https://www.youtube.com/embed/${videoId}?list=${playlistId}`
                      : `https://www.youtube.com/embed/${videoId}`,
                  }
                } else if (playlistId) {
                  // YouTube playlist only.
                  node.tagName = "iframe"
                  node.properties = {
                    class: "external-embed",
                    allow: "fullscreen",
                    frameborder: 0,
                    width: "600px",
                    height: "350px",
                    src: `https://www.youtube.com/embed/videoseries?list=${playlistId}`,
                  }
                }
              }