Website comments

11 Aug 2019

Today I am working on adding functionality on my website for users to add comments. some people have already done this, but I decided to not go with their way and just use the bare-bones github issues functionality without much interfacing.

Unlike most previous posts, I am writing this post while actually working on the project (instead of after I’m finished.). Hopefully this goes well.

The first thing I just did was add a piece of metadata to each existing post. I called it comment_id . I started manually adding it to each of my posts, but then I thought, why am I doing this manually with VIM in a linux terminal? I should be able to at least automate part of it. So, I did some research into using SED to do what I want. Before adding comment_id, my post headers looked like this:

layout: post
title "Website comments"
date: 2019-08-11

What I wanted to do was to reguar-expression-match on date and insert a line below, like so:

layout: post
title: "Website comments"
daet: 2019-08-11
comment_id: 9

After looking around here, I sort of got an idea of what I wanted to do:

sed -i.bak 'N;/date: [0-9-]\+.*\n/s/\(date: [0-9-]\+\)/\1\ncomment_id: 8/;P;D'

The key parts are:

  1. N;/date: [0-9-]\+.*\n/ : Append (N) the next line for operationsif the current line matches date: [0-9-]\+.*\n
  2. s/\(date: [0-9-]\+\)/\1\ncomment_id: 8/; : Replace occurrences of \(date: [0-9-]\+\) with \1\ncomment_id: 8
  3. P;D : print everything up to the first new line in pattern space; delete everything up to first new line in pattern space
  4. -i.bak, : do this on and create backups with .bak suffix

That ended up working great. I definitely could have automated the numbering process even more with a bash FOR loop or some sed-awk stuff I think, but didn’t want to go that far. I eventually decided to use post_id instead of comment_id, so I ran sed -i.bak s/comment_id/post_id/ *.md on my files (NOTE: If I ever do this again, I will have to be more specific in my expression, since I reference comment_id and post_id several times in this post).

Once I had done that, I needed to go and manually make an issue for every website post so far. Hopefully this is something I can automate with a github bot in the future. I also needed to slightly modify my _posts layout to add some text at the botton of each page with a link to the relevant issues page. If I did my job correctly, it should appear as the next sentence.

Have feedback on this post? Comment Here