Post - rebundling and the script for it

content/posts/2023/10/big-rebundle/index.md

@@ -0,0 +1,79 @@
+---
+title: "The big rebundle"
+description: "Aided by an AI-generated script, I undo my hasty action from earlier this year."
+author: Bryce Wray
+date: 2023-10-06T12:08:00-05:00
+# draft: true
+# initTextEditor: iA Writer # default --- change if needed
+# imgs:
+# -
+---
+
+I'll cut to the chase: after an [admittedly rash decision](/posts/2023/07/big-unbundle/) of a few months ago, I've reverted to an [earlier choice](/posts/2022/07/bundling-up-rebuilding-my-hugo-site) in favor of using [page bundles](https://gohugo.io/content-management/page-bundles/) for managing this [Hugo](https://gohugo.io)-based website.
+
+But how did I handle that without a whole lot of manual agony? That's the real story of this post because, although I'm [on record](/posts/2023/03/no-ai-here-thanks-very-much/) as not being in favor of AI-generated *content*, an AI-generated *script* saved me from hours of trouble.
+
+<!--more-->
+
+I started in on changing my content back to the bundled setup --- over 300 posts, one frickin’ post at a time. After getting a few posts into the process, I stopped briefly to review a post by fellow Hugo user [Pawel Grzybek](https://github.com/pawelgrzybek), "[Migration from single content files to Hugo page bundles](https://pawelgrzybek.com/migration-from-single-content-files-to-hugo-page-bundles/)," which had been instrumental in convincing me to rebundle the site. Down in the post's comments section, a reader asked how long the operation had taken. Grzybek's reply, in part:
+
+> It was actually a very quick process. I wrote a small Node.js script that moved all the files to a new location for me. Writing the script didn't take me longer than 30 minutes, and the script execution took only a few hundred milliseconds.
+
+"A *script*," I muttered, doing a mental facepalm, and added to myself: "--- of *course*, you moron."
+
+Only thing was, given my own limited skill set, I knew it would take *me* quite a while to write a script to handle this for my purposes. In fact, I figured, I could manually complete all the necessary file renames and moves well before I could arrive at a script that not only did them for me but also *didn't* eat my files in the process. So I went the lazy route and asked [Google's Bard AI product](https://bard.google.com) to write me a Bash script to accomplish what I wanted. To my surprise, given how skeptical I've become of AI's ability to do things without "hallucinating," Bard produced *exactly* what I wanted.
+
+I first prompted Bard with:
+
+> Write a Bash script that will look through directories and move each .md file into a subfolder with the same name (except for the .md extension), then move the .md file into the subfolder and rename the .md file "index.md" — for example, if it finds /content/posts/2018/09/my-post.md, it should create /content/posts/2018/09/my-post/, then `mv` the .md file to /content/posts/2018/09/my-post/index.md. It should do this recursively.
+
+It responded with one that would do that, but would have to be moved into each directory to handle the moves for that directory. That was my fault; I hadn't been sufficiently specific about what I wanted. So, then, I further prompted:
+
+> Alter the Bash script so that it prompts me for the directory and then does the same thing I've described, so I don't have to manually move it into directories.
+
+. . . and Bard provided this final script, which worked like a dream after I assigned it the proper permissions[^permissions]:
+
+[^permissions]: `chmod +x move-to-bundles.sh` --- and Bard even reminded me to do that.
+
+```bash{filename="move-to-bundles.sh" bigdiv=true}
+#!/bin/bash
+
+# Prompt the user for the directory to scan
+echo "Enter the directory to scan:"
+read dir
+
+# Check if the directory exists
+if [ ! -d "$dir" ]; then
+	echo "Directory does not exist."
+	exit 1
+fi
+
+# Change to the specified directory
+cd "$dir"
+
+# Recursively find all .md files in the current directory and subdirectories
+for file in $(find . -name "*.md"); do
+
+	# Get the directory containing the file
+	dir=$(dirname "$file")
+	
+	# Get the filename without the .md extension
+	filename=$(basename "$file" .md)
+	
+	# Create a subfolder with the same name as the file, if it doesn't exist yet
+	if [ ! -d "$dir/$filename" ]; then
+		mkdir "$dir/$filename"
+	fi
+	
+	# Move the .md file to the subfolder and rename it "index.md"
+	mv "$file" "$dir/$filename/index.md"
+done
+```
+
+With this, I simply ran the script within the site repo and, when prompted, specified (*e.g.*) `content/posts/2019/` --- I went one year at a time, to minimize the potential chaos if it *didn't* do what I wanted --- and let the script handle the rest.
+
+Now, of course, this didn't do *everything* I needed; I still had to move related image files into the same folders with those content files that called them.[^graphics] However, the script had taken care of the real grunt work, so moving the image files was a piece of cake compared to what I'd otherwise have endured.
+
+[^graphics]: I also had to revert the minor graphics-related code changes to which I referred in a footnote to "[The big unbundle](/posts/2023/07/big-unbundle/)." (Thank goodness for Git.)
+
+I suspect going back in the other direction --- *i.e.*, unbundling the site once again --- wouldn't be as neatly scriptable as the rebundling proved to be. Perhaps that'll provide a sufficiently strong incentive for me *not* to futz with this again.