Jaspers Research Agents: Agents you define that answer research chats from SEC filings, your documents, and the web, citing every passage.
| 1 | import type { Definition, Root, RootContent } from 'mdast' |
| 2 | import { fromMarkdown } from 'mdast-util-from-markdown' |
| 3 | import { gfmFromMarkdown } from 'mdast-util-gfm' |
| 4 | import { gfm } from 'micromark-extension-gfm' |
| 5 | |
| 6 | // An answer's Markdown read into a tree, which the chat draws as React elements, never as HTML, the |
| 7 | // way the markdown plugin draws a document: GitHub-flavored, links https only, HTML as its text |
| 8 | // except <br>. Pure, so the reading can be checked in a test. |
| 9 | |
| 10 | /** A <br> models put in a table cell, or anywhere, for a line break. */ |
| 11 | const LINE_BREAK = /^<br\s*\/?>$/i |
| 12 | |
| 13 | export function parseMarkdown(source: string): Root { |
| 14 | return fromMarkdown(source, { extensions: [gfm()], mdastExtensions: [gfmFromMarkdown()] }) |
| 15 | } |
| 16 | |
| 17 | /** The address a link opens, or null: main opens https links only, so nothing else is a link here. */ |
| 18 | export function linkHref(url: string): string | null { |
| 19 | try { |
| 20 | return new URL(url).protocol === 'https:' ? url : null |
| 21 | } catch { |
| 22 | return null |
| 23 | } |
| 24 | } |
| 25 | |
| 26 | export function isLineBreak(html: string): boolean { |
| 27 | return LINE_BREAK.test(html) |
| 28 | } |
| 29 | |
| 30 | /** The definitions reference links point at, by the label mdast gives both, wherever they sit. */ |
| 31 | export function definitionsOf(tree: Root): Map<string, Definition> { |
| 32 | const found = new Map<string, Definition>() |
| 33 | const visit = (nodes: RootContent[]): void => { |
| 34 | for (const node of nodes) { |
| 35 | if (node.type === 'definition') { |
| 36 | if (!found.has(node.identifier)) found.set(node.identifier, node) |
| 37 | } else if ('children' in node) visit(node.children) |
| 38 | } |
| 39 | } |
| 40 | visit(tree.children) |
| 41 | return found |
| 42 | } |