eng-1213-adjustments to markdown: use roam links

apps/roam/src/utils/jsonld.ts

@@ -108,10 +108,12 @@ export const getJsonLdData = async ({
   Record<string, string | Record<string, string> | Record<string, string>[]>
 > => {
   const roamUrl = canonicalRoamUrl();
+  const roamUrlRe = new RegExp(roamUrl + "/page/([\\w\\d-]{9,10})\\b", "g");
   const getRelationData = () =>
     getRelationDataUtil({ allRelations, nodeLabelByType });
   await updateExportProgress(0);
   const pageData = getPageData({ results, allNodes });
+  const nodeUids = new Set(pageData.map((p) => p.uid));
   const numPages = pageData.length + allNodes.length;
   let numTreatedPages = 0;
   const settings = {
@@ -138,6 +140,8 @@ export const getJsonLdData = async ({
         ...settings,
         allNodes,
         linkType: "roam url",
+        blockAnchors: true,
+        blockRefsAsLinks: true,
       });
       page.content = r.content;
       numTreatedPages += 1;
@@ -153,6 +157,11 @@ export const getJsonLdData = async ({
         error: `Unknown node type "${type}" for page "${text}"`,
       });
     }
+    const textRefersToNode = [
+      ...new Set([...(content as string).matchAll(roamUrlRe)].map((r) => r[1])),
+    ]
+      .filter((r) => nodeUids.has(r))
+      .map((r) => `pages:${r}`);
     const r = {
       "@id": `pages:${uid}`, // eslint-disable-line @typescript-eslint/naming-convention
       "@type": nodeType ?? "nodeSchema", // eslint-disable-line @typescript-eslint/naming-convention
@@ -162,6 +171,7 @@ export const getJsonLdData = async ({
       created: date.toJSON(),
       creator: displayName,
     };
+    if (textRefersToNode.length > 0) return { ...r, textRefersToNode };
     return r;
   });
   const nodeSet = new Set(pageData.map((n) => n.uid));

apps/roam/src/utils/pageToMarkdown.ts

@@ -1,5 +1,6 @@
 import { BLOCK_REF_REGEX } from "roamjs-components/dom/constants";
 import normalizePageTitle from "roamjs-components/queries/normalizePageTitle";
+import getPageUidByBlockUid from "roamjs-components/queries/getPageUidByBlockUid";
 import getTextByBlockUid from "roamjs-components/queries/getTextByBlockUid";
 import getPageMetadata from "./getPageMetadata";
 import getPageViewType from "roamjs-components/queries/getPageViewType";
@@ -20,6 +21,7 @@ import {
   pullBlockToTreeNode,
   collectUids,
 } from "./exportUtils";
+import getPageUidByPageTitle from "roamjs-components/queries/getPageUidByPageTitle";
 
 const MATCHES_NONE = /$.+^/;
 
@@ -127,6 +129,8 @@ export const toMarkdown = ({
     removeSpecialCharacters: boolean;
     linkType: string;
     flatten?: boolean;
+    blockRefsAsLinks?: boolean;
+    blockAnchors?: boolean;
   };
 }): string => {
   const {
@@ -138,35 +142,53 @@ export const toMarkdown = ({
     removeSpecialCharacters,
     linkType,
     flatten = false,
+    blockRefsAsLinks = false,
+    blockAnchors = false,
   } = opts;
   const processedText = c.text
-    .replace(embeds ? EMBED_REGEX : MATCHES_NONE, (_, blockUid) => {
+    .replace(embeds ? EMBED_REGEX : MATCHES_NONE, (_, blockUid: string) => {
       const reference = getFullTreeByParentUid(blockUid);
       return toMarkdown({ c: reference, i, v, opts });
     })
-    .replace(embeds ? EMBED_CHILDREN_REGEX : MATCHES_NONE, (_, blockUid) => {
-      const reference = getFullTreeByParentUid(blockUid);
-      return reference.children
-        .map((child) => toMarkdown({ c: child, i, v, opts }))
-        .join("\n");
-    })
-    .replace(refs ? BLOCK_REF_REGEX : MATCHES_NONE, (_, blockUid) => {
-      const reference = getTextByBlockUid(blockUid);
-      return reference || blockUid;
+    .replace(
+      embeds ? EMBED_CHILDREN_REGEX : MATCHES_NONE,
+      (_, blockUid: string) => {
+        const reference = getFullTreeByParentUid(blockUid);
+        return reference.children
+          .map((child) => toMarkdown({ c: child, i, v, opts }))
+          .join("\n");
+      },
+    )
+    .replace(refs ? BLOCK_REF_REGEX : MATCHES_NONE, (_, blockUid: string) => {
+      const reference = getTextByBlockUid(blockUid) || blockUid;
+
+      if (blockRefsAsLinks) {
+        const pageUid = blockAnchors
+          ? getPageUidByBlockUid(blockUid)
+          : blockUid;
+        if (pageUid === blockUid || pageUid === "")
+          return toLink(reference, blockUid, linkType);
+        else
+          // Note that the roam anchor follows a more complex pattern
+          // Here using the block anchor as target for internal export consistency
+          return toLink(reference, `${pageUid}#block-${blockUid}`, linkType);
+      } else return reference;
     })
     .replace(/{{\[\[TODO\]\]}}/g, v === "bullet" ? "[ ]" : "- [ ]")
     .replace(/{{\[\[DONE\]\]}}/g, v === "bullet" ? "[x]" : "- [x]")
     .replace(/\_\_(.+?)\_\_/g, "_$1_") // convert Roam italics __ to markdown italics _
     .replace(/(?<!\n)```/g, "\n```") // Add line break before last code blocks
     .trim();
   const finalProcessedText =
-    simplifiedFilename || removeSpecialCharacters
+    simplifiedFilename || removeSpecialCharacters || blockRefsAsLinks
       ? XRegExp.matchRecursive(processedText, "#?\\[\\[", "\\]\\]", "i", {
           valueNames: ["between", "left", "match", "right"],
           unbalanced: "skip",
         })
           .map((s) => {
             if (s.name === "match") {
+              const pageUid = getPageUidByPageTitle(s.value);
+              if (pageUid.length > 0) return toLink(s.value, pageUid, linkType);
               const name = getFilename({
                 title: s.value,
                 allNodes,
@@ -205,8 +227,8 @@ export const toMarkdown = ({
     })
     .join("");
   const lineBreak = v === "document" ? "\n" : "";
-
-  return `${indentation}${viewTypePrefix}${headingPrefix}${finalProcessedText}${lineBreak}${childrenMarkdown}`;
+  const blockAnchor = blockAnchors ? `{#block-${c.uid}}` : "";
+  return `${indentation}${viewTypePrefix}${headingPrefix}${blockAnchor}${finalProcessedText}${lineBreak}${childrenMarkdown}`;
 };
 
 export const pageToMarkdown = async (
@@ -222,6 +244,8 @@ export const pageToMarkdown = async (
     maxFilenameLength,
     removeSpecialCharacters,
     linkType,
+    blockRefsAsLinks = false,
+    blockAnchors = false,
   }: {
     includeDiscourseContext: boolean;
     appendRefNodeContext: boolean;
@@ -233,6 +257,8 @@ export const pageToMarkdown = async (
     maxFilenameLength: number;
     removeSpecialCharacters: boolean;
     linkType: string;
+    blockRefsAsLinks?: boolean;
+    blockAnchors?: boolean;
   },
 ): Promise<{ title: string; content: string; uids: Set<string> }> => {
   const v = getPageViewType(text) || "bullet";
@@ -287,6 +313,8 @@ export const pageToMarkdown = async (
           maxFilenameLength,
           removeSpecialCharacters,
           linkType,
+          blockAnchors,
+          blockRefsAsLinks,
         },
       }),
     )