From 50105fd044f4488902404dded9fa23e0b143933c Mon Sep 17 00:00:00 2001
From: Ian Shehadeh <ian@shehadeh.net>
Date: Thu, 12 Oct 2023 14:30:47 -0400
Subject: [PATCH] Index plug: index paragraph nodes (#528)

* index plug: index paragraph nodes

Includes attributes and tags in the paragraphs.
 Only considers top level paragraphs (right below Document node)

* keep ref and pos consistent with other objects

- most objects just store the starting position in the `pos` field.
- ref is usually `${page}@${pos}` so that it works as a link.

* cleanup and clarify comments

* add paragraph to builtins index
---
 plugs/index/builtins.ts     |  5 ++++
 plugs/index/index.plug.yaml |  5 ++++
 plugs/index/paragraph.ts    | 51 +++++++++++++++++++++++++++++++++++++
 3 files changed, 61 insertions(+)
 create mode 100644 plugs/index/paragraph.ts

diff --git a/plugs/index/builtins.ts b/plugs/index/builtins.ts
index 49fe15c0..7eaa7507 100644
--- a/plugs/index/builtins.ts
+++ b/plugs/index/builtins.ts
@@ -53,6 +53,11 @@ export const builtins: Record<string, Record<string, string>> = {
     inDirective: "boolean",
     asTemplate: "boolean",
   },
+  paragraph: {
+    text: "string",
+    page: "string",
+    pos: "number",
+  },
 };
 
 export async function loadBuiltinsIntoIndex() {
diff --git a/plugs/index/index.plug.yaml b/plugs/index/index.plug.yaml
index 488da23a..ff5f1f32 100644
--- a/plugs/index/index.plug.yaml
+++ b/plugs/index/index.plug.yaml
@@ -86,6 +86,11 @@ functions:
     events:
       - page:index
 
+  indexParagraphs:
+    path: "./paragraph.ts:indexParagraphs"
+    events:
+      - page:index
+
   # Backlinks
   indexLinks:
     path: "./page_links.ts:indexLinks"
diff --git a/plugs/index/paragraph.ts b/plugs/index/paragraph.ts
new file mode 100644
index 00000000..9619e681
--- /dev/null
+++ b/plugs/index/paragraph.ts
@@ -0,0 +1,51 @@
+import type { IndexTreeEvent } from "$sb/app_event.ts";
+import { indexObjects } from "./api.ts";
+import { renderToText, traverseTree, traverseTreeAsync } from "$sb/lib/tree.ts";
+import { extractAttributes } from "$sb/lib/attribute.ts";
+
+/** ParagraphObject  An index object for the top level text nodes */
+export type ParagraphObject = {
+  ref: string;
+  tags: string[];
+  text: string;
+  page: string;
+  pos: number;
+} & Record<string, any>;
+
+export async function indexParagraphs({ name: page, tree }: IndexTreeEvent) {
+  const objects: ParagraphObject[] = [];
+
+  await traverseTreeAsync(tree, async (p) => {
+    // only search directly under document
+    //  Paragraph nodes also appear under block elements
+    if (p.type == "Document") return false; // continue traversal if p is Document
+    if (p.type != "Paragraph") return true;
+
+    const tags = new Set<string>(["paragraph"]);
+    // tag the paragraph with any hashtags inside it
+    traverseTree(p, (e) => {
+      if (e.type == "Hashtag") {
+        tags.add(e.children![0].text!.substring(1));
+        return true;
+      }
+
+      return false;
+    });
+
+    const attrs = await extractAttributes(p, false);
+    const pos = p.from!;
+    objects.push({
+      ref: `${page}@${pos}`,
+      text: renderToText(p),
+      tags: [...tags.values()],
+      page,
+      pos,
+      ...attrs,
+    });
+
+    // stop on every element except document, including paragraphs
+    return true;
+  });
+
+  await indexObjects<ParagraphObject>(page, objects);
+}