@mintlify/scraping 4.0.1168 → 4.0.1170

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mintlify/scraping",
-  "version": "4.0.1168",
+  "version": "4.0.1170",
   "description": "Scrape documentation frameworks to Mintlify docs",
   "engines": {
     "node": ">=18.0.0"
@@ -43,7 +43,7 @@
     "format:check": "oxfmt --check"
   },
   "dependencies": {
-    "@mintlify/common": "1.0.1295",
+    "@mintlify/common": "1.0.1296",
     "fast-xml-parser": "5.7.3",
     "fs-extra": "11.1.1",
     "graphql": "16.11.0",
@@ -87,5 +87,5 @@
       "optional": true
     }
   },
-  "gitHead": "2b6f6889e965e0aeb88757feb55e69daa4255d37"
+  "gitHead": "21e4cfc701552fe0fb6d40c8b5b4896fb50c08ff"
 }

package/src/sdk/convert.ts

@@ -1,5 +1,6 @@
 import type { Root as HastRoot, Element } from 'hast';
 import type {
+  Blockquote,
   Definition,
   Heading,
   Html,
@@ -12,6 +13,7 @@ import type {
   Root as MdastRoot,
   Text,
 } from 'mdast';
+import type { MdxJsxFlowElement } from 'mdast-util-mdx-jsx';
 import remarkGfm from 'remark-gfm';
 import remarkMdx from 'remark-mdx';
 import remarkParse from 'remark-parse';
@@ -99,6 +101,40 @@ export function htmlToMdx(html: string, rewriteLink?: RewriteLink): string {
 // paragraph and keep their block boundaries; inline HTML becomes text within
 // its paragraph, merged with its neighbours so a paragraph that opens with a
 // tag still starts with its first word.
+const GITHUB_ALERT_CALLOUTS: Record<string, string> = {
+  NOTE: 'Note',
+  TIP: 'Tip',
+  IMPORTANT: 'Info',
+  WARNING: 'Warning',
+  CAUTION: 'Danger',
+};
+const GITHUB_ALERT = /^\[!(NOTE|TIP|IMPORTANT|WARNING|CAUTION)\][ \t]*(\n|$)/i;
+
+function githubAlertsToCallouts(tree: MdastRoot): void {
+  visit(tree, 'blockquote', (node: Blockquote, index, parent) => {
+    if (!parent || index === undefined) return;
+    const [first] = node.children;
+    if (first?.type !== 'paragraph') return;
+    const [lead] = first.children;
+    if (lead?.type !== 'text') return;
+    const match = GITHUB_ALERT.exec(lead.value);
+    const name = match?.[1] ? GITHUB_ALERT_CALLOUTS[match[1].toUpperCase()] : undefined;
+    if (!match || !name) return;
+    const next = first.children[1];
+    if (match[2] === '' && next !== undefined && next.type !== 'break') return;
+    lead.value = lead.value.slice(match[0].length);
+    if (lead.value === '') first.children.shift();
+    if (first.children[0]?.type === 'break') first.children.shift();
+    const callout: MdxJsxFlowElement = {
+      type: 'mdxJsxFlowElement',
+      name,
+      attributes: [],
+      children: first.children.length > 0 ? node.children : node.children.slice(1),
+    };
+    parent.children.splice(index, 1, callout);
+  });
+}
+
 export function parseMarkdown(markdown: string, minHeadingDepth?: Heading['depth']): MdastRoot {
   // CommonMark treats \r\n and a bare \r as line endings; normalising them up
   // front means every later split and check only has to know about \n.
@@ -132,6 +168,7 @@ export function parseMarkdown(markdown: string, minHeadingDepth?: Heading['depth
   visit(mdastTree, 'paragraph', (node: Paragraph) => {
     node.children = mergeText(node.children);
   });
+  githubAlertsToCallouts(mdastTree);
   if (minHeadingDepth) {
     visit(mdastTree, 'heading', (node: Heading) => {
       node.depth = node.depth < minHeadingDepth ? minHeadingDepth : node.depth;

package/src/sdk/converters/go.ts

@@ -404,7 +404,7 @@ class GoRenderer {
   private packagePage(): SdkPage {
     const slug =
       this.index.packageSlug(this.pkg.importPath) ?? path.posix.join(this.base, PACKAGE_PAGE);
-    const lines = [this.prose(this.pkg.doc), this.deprecated(this.pkg.deprecated)];
+    const lines = [this.body(this.pkg.doc), this.deprecated(this.pkg.deprecated)];
     const types = this.pkg.types ?? [];
     if (types.length) lines.push('## Types', this.indexList(types.map((type) => type.name)));
     if (this.funcs.length) {
@@ -505,7 +505,7 @@ class GoRenderer {
 
   private typePage(type: GoType): SdkPage {
     const lines = [
-      this.prose(type.doc),
+      this.body(type.doc),
       this.deprecated(type.deprecated),
       codeBlock(declarationHeader(type)),
     ];
@@ -535,7 +535,7 @@ class GoRenderer {
 
   private funcPage(fn: GoFunc): SdkPage {
     const lines = [
-      this.prose(fn.doc),
+      this.body(fn.doc),
       this.deprecated(fn.deprecated),
       codeBlock(codeText(fn.decl)),
       ...this.signatureDetails(fn, 2),
@@ -736,6 +736,24 @@ class GoRenderer {
     return markdown ? stringifyMdast(this.doc(markdown).tree) : '';
   }
 
+  private body(markdown?: string): string {
+    if (!markdown) return '';
+    const { tree } = this.doc(markdown);
+    const [first] = tree.children;
+    if (first?.type !== 'paragraph') return stringifyMdast(tree);
+    const text = plainText(first).replace(/\s+/g, ' ').trim();
+    const sentence = leadingSentence(text);
+    if (!sentence || sentence === text || truncate(sentence, MAX_DESCRIPTION) !== sentence) {
+      return stringifyMdast(tree);
+    }
+    const copy = structuredClone(tree);
+    const [paragraph] = copy.children;
+    if (paragraph?.type !== 'paragraph') return stringifyMdast(tree);
+    dropLeadingSentence(paragraph);
+    if (paragraph.children.length === 0) copy.children.shift();
+    return stringifyMdast(copy);
+  }
+
   private doc(markdown: string): GoDoc {
     let doc = this.rendered.get(markdown);
     if (!doc) {