From 85d32f3ed50f8bdde4b89cbbf5475ed485ff9ee9 Mon Sep 17 00:00:00 2001
From: Cristian Lara <cristianlara@meta.com>
Date: Mon, 2 Dec 2024 17:15:29 -0500
Subject: [PATCH] [Tutorials] Improve conversion of math to MDX

- Properly convert /begin.../end math blocks by wrapping in $$
- Update our regex for unescaping braces in math to support math wrapped by double dollar signs and containing newlines.
- Make sure $$ symbols are not escaped and include line breaks before and after. I encountered a few examples in our notebooks that didn't match this formatting expected by mdx.

These changes were made when adapting this script to work in the botorch repo:
- https://github.com/CristianLara/botorch/pull/3/commits/a0b3eaac561d1a78dc5015e938a10c44e5cfbe39
- https://github.com/CristianLara/botorch/pull/3/commits/5118c533766f7d7c958e61c6fb37a3fb892e2819
---
 scripts/convert_ipynb_to_mdx.py | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/scripts/convert_ipynb_to_mdx.py b/scripts/convert_ipynb_to_mdx.py
index b98151f05c1..37aec12bbcd 100644
--- a/scripts/convert_ipynb_to_mdx.py
+++ b/scripts/convert_ipynb_to_mdx.py
@@ -312,13 +312,20 @@ def sanitize_mdx(mdx: str) -> str:
     # Remove some lingering HTML tags that break MDX.
     mdx = mdx.replace("<markdowncell>", "")
     mdx = mdx.replace("<TODO>", "")
+    mdx = mdx.replace("<br>", "<br />")
     # Remove any HTML comments from the Markdown. They are fine to keep in the
     # notebooks, but are not really useful in the MDX.
     mdx = re.sub("(<!--.*?-->)", "", mdx, flags=re.DOTALL)
     # "\" Escape braces to make the text MDX compatible.
     mdx = re.sub("([^\\\\])([{}])", "\\g<1>\\\\\\g<2>", mdx)
-    # Escaping braces causes issues in math blocks, unescape them.
-    mdx = re.sub("\\$(.*?)\\$", lambda match: match[0].replace("\\{", "{").replace("\\}", "}"), mdx)
+
+    # -- KaTeX --
+    # Wrap '\begin{}...\end{}' in $$ for KaTeX to work.
+    mdx = re.sub("(\\\\begin\\\\{(\\w*?)\\\\}(.|\n)*?end\\\\{\\2\\\\})", "$$\\g<1>$$", mdx)
+    # # make sure $$ symbols are not escaped and include line breaks.
+    mdx = re.sub("\\\\?\\$\\\\?\\$((?:.|\n)*?)\\\\?\\$\\\\?\\$", "\n$$\n\\g<1>\n$$\n", mdx)
+    # # Escaping braces causes issues in math blocks, unescape them.
+    mdx = re.sub("\\$?\\$(.|\n)*?\\$\\$?", lambda match: match[0].replace("\\{", "{").replace("\\}", "}"), mdx)
 
     return mdx