fix scanning for tags in content

Jacky Zhao

2023-08-12 83269ac26e41f8e81b56e000b4347ee60d4ff84b

 quartz/plugins/transformers/links.ts

@@ -1,117 +1,122 @@
import { QuartzTransformerPlugin } from "../types"
import { CanonicalSlug, transformInternalLink } from "../../path"
import {
  CanonicalSlug,
  RelativeURL,
  _stripSlashes,
  canonicalizeServer,
  joinSegments,
  pathToRoot,
  resolveRelative,
  splitAnchor,
  transformInternalLink,
} from "../../path"
import path from "path"
import { visit } from 'unist-util-visit'
import { visit } from "unist-util-visit"
import isAbsoluteUrl from "is-absolute-url"

interface Options {
  /** How to resolve Markdown paths */
  markdownLinkResolution: 'absolute' | 'relative' | 'shortest'
  markdownLinkResolution: "absolute" | "relative" | "shortest"
  /** Strips folders from a link so that it looks nice */
  prettyLinks: boolean
  indexAnchorLinks: boolean
  indexExternalLinks: boolean
}

const defaultOptions: Options = {
  markdownLinkResolution: 'absolute',
  markdownLinkResolution: "absolute",
  prettyLinks: true,
  indexAnchorLinks: false,
  indexExternalLinks: false,
}

export const CrawlLinks: QuartzTransformerPlugin<Partial<Options> | undefined> = (userOpts) => {
  const opts = { ...defaultOptions, ...userOpts }
  return {
    name: "LinkProcessing",
    htmlPlugins() {
      return [() => {
        return (tree, file) => {
          const curSlug = file.data.slug!
          const transformLink = (target: string) => {
            const targetSlug = transformInternalLink(target) 
            if (opts.markdownLinkResolution === 'relative' && !path.isAbsolute(targetSlug)) {
              return './' + relative(curSlug, targetSlug)
            } else if (opts.markdownLinkResolution === 'shortest') {
              // https://forum.obsidian.md/t/settings-new-link-format-what-is-shortest-path-when-possible/6748/5
              const allSlugs = file.data.allSlugs!
    htmlPlugins(ctx) {
      return [
        () => {
          return (tree, file) => {
            const curSlug = canonicalizeServer(file.data.slug!)
            const transformLink = (target: string): RelativeURL => {
              const targetSlug = _stripSlashes(transformInternalLink(target).slice(".".length))
              let [targetCanonical, targetAnchor] = splitAnchor(targetSlug)
              if (opts.markdownLinkResolution === "relative") {
                return targetSlug as RelativeURL
              } else if (opts.markdownLinkResolution === "shortest") {
                // if the file name is unique, then it's just the filename
                const matchingFileNames = ctx.allSlugs.filter((slug) => {
                  const parts = slug.split(path.posix.sep)
                  const fileName = parts.at(-1)
                  return targetCanonical === fileName
                })

              // if the file name is unique, then it's just the filename
              const matchingFileNames = allSlugs.filter(slug => {
                const parts = toServerSlug(slug).split(path.posix.sep)
                const fileName = parts.at(-1)
                return targetSlug === fileName
              })
                // only match, just use it
                if (matchingFileNames.length === 1) {
                  const targetSlug = canonicalizeServer(matchingFileNames[0])
                  return (resolveRelative(curSlug, targetSlug) + targetAnchor) as RelativeURL
                }

              if (matchingFileNames.length === 1) {
                const targetSlug = toServerSlug(matchingFileNames[0])
                return './' + relativeToRoot(curSlug, targetSlug)
                // if it's not unique, then it's the absolute path from the vault root
                // (fall-through case)
              }

              // if it's not unique, then it's the absolute path from the vault root
              // (fall-through case)
              // treat as absolute
              return joinSegments(pathToRoot(curSlug), targetSlug) as RelativeURL
            }
            // treat as absolute
            return './' + relativeToRoot(curSlug, targetSlug)

            const outgoing: Set<CanonicalSlug> = new Set()
            visit(tree, "element", (node, _index, _parent) => {
              // rewrite all links
              if (
                node.tagName === "a" &&
                node.properties &&
                typeof node.properties.href === "string"
              ) {
                let dest = node.properties.href as RelativeURL
                node.properties.className ??= []
                node.properties.className.push(isAbsoluteUrl(dest) ? "external" : "internal")

                // don't process external links or intra-document anchors
                if (!(isAbsoluteUrl(dest) || dest.startsWith("#"))) {
                  dest = node.properties.href = transformLink(dest)
                  const canonicalDest = path.posix.normalize(joinSegments(curSlug, dest))
                  const [destCanonical, _destAnchor] = splitAnchor(canonicalDest)
                  outgoing.add(destCanonical as CanonicalSlug)
                }

                // rewrite link internals if prettylinks is on
                if (
                  opts.prettyLinks &&
                  node.children.length === 1 &&
                  node.children[0].type === "text" &&
                  !node.children[0].value.startsWith("#")
                ) {
                  node.children[0].value = path.basename(node.children[0].value)
                }
              }

              // transform all other resources that may use links
              if (
                ["img", "video", "audio", "iframe"].includes(node.tagName) &&
                node.properties &&
                typeof node.properties.src === "string"
              ) {
                if (!isAbsoluteUrl(node.properties.src)) {
                  let dest = node.properties.src as RelativeURL
                  const ext = path.extname(node.properties.src)
                  dest = node.properties.src = transformLink(dest)
                  node.properties.src = dest + ext
                }
              }
            })

            file.data.links = [...outgoing]
          }

          const outgoing: Set<CanonicalSlug> = new Set()
          visit(tree, 'element', (node, _index, _parent) => {
            // rewrite all links
            if (
              node.tagName === 'a' &&
              node.properties &&
              typeof node.properties.href === 'string'
            ) {
              let dest = node.properties.href
              node.properties.className = isAbsoluteUrl(dest) ? "external" : "internal"

              // don't process external links or intra-document anchors
              if (!(isAbsoluteUrl(dest) || dest.startsWith("#"))) {
                node.properties.href = transformLink(dest)
              }

              dest = node.properties.href
              if (dest.startsWith(".")) {
                const normalizedPath = path.normalize(path.join(curSlug, dest))
                outgoing.add(trimPathSuffix(normalizedPath))
              } else if (dest.startsWith("#")) {
                if (opts.indexAnchorLinks) {
                  outgoing.add(dest)
                }
              } else {
                if (opts.indexExternalLinks) {
                  outgoing.add(dest)
                }
              }

              // rewrite link internals if prettylinks is on
              if (opts.prettyLinks && node.children.length === 1 && node.children[0].type === 'text') {
                node.children[0].value = path.basename(node.children[0].value)
              }
            }

            // transform all other resources that may use links
            if (
              ["img", "video", "audio", "iframe"].includes(node.tagName) &&
              node.properties &&
              typeof node.properties.src === 'string'
            ) {
              if (!isAbsoluteUrl(node.properties.src)) {
                const ext = path.extname(node.properties.src)
                node.properties.src = transformLink(path.join("assets", node.properties.src)) + ext
              }
            }
          })

          file.data.links = [...outgoing]
        }
      }]
    }
        },
      ]
    },
  }
}

declare module 'vfile' {
declare module "vfile" {
  interface DataMap {
    links: CanonicalSlug[]
  }