Note: I don’t use Clapper’s image popup plugin anymore, so the settings below are now gone.

Background

Using the following plugin, I finished removing the repeating figures in my Atom Feed.¹

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

module CustomLiquidFilters
  def remove_linenumbers(input)
    input.gsub(/\<td\ class="gutter"\>.+?\<\/td\>/m, ' ')
  end

  RegexMap = {
    /\<div\ class="caption"\>Click the image for a larger view.\<\/div\>/ => '',
    /\<div\ id="image-dialog-\d+".*?\>.+?\<\/div\>/m => '',
    /\<div\ class="illustration\ print"\>.+?\<\/div\>/m => '',
  }

  def remove_bigfig(input)
    RegexMap.each_pair {|k,v| input.gsub!(k, v)}
    input
  end
end

Liquid::Template.register_filter CustomLiquidFilters

Problem

How about category Atom Feeds?

It took me some time a find out that the file needed to be changed was source/_includes/custom/category_feed.xml. In that file, I tried to do the same modification like what I had already did in source/atom.xml.

1
2
3
4

<entry>
  <!-- other elements -->
  <content type="html"><![CDATA[{{ post.content | remove_linenumbers | remove_bigfig | expand_urls: site.url | markdownify | cdata_escape }}]]></content>
</entry>

Then I generated the site without getting any errors. However, when I was previewing one of my category Atom Feeds, I found out that the above changes to the XML had did nothing—the popup images still appeared twice.

Exploring the default Jekyll code

I tried deleting my custom filter remove_bigfig, so that I could test whether Walter’s instructions for removing line numbers worked for category Atom Feeds. Unfortunately, his method failed.

Having no knowledge in how the XML file actually worked, I broke things down by deleting the word markdownify, as well as remove_bigfig, in order to see how things worked. Then I saw markdown code at localhost:4000. It means that markdownify converts markdown to HTML. Thus, I tried to insert the word remove_bigfig again just after markdownify, but I still saw the same popup image appearing twice.

Since Walter’s steps didn’t work for me, I almost wanted to give it up.

In the morning two days ago, I suddenly realized that I could write see what’s going on inside Octopress by getting the standard output. Not going take an online tutorial about using Ruby debugger, I used the most primitive way of debugging.

1
2
3

def debug(input)
  puts input
end

I recorded the output into a file and then browsed through it. I found the imgpopup ... inside the file. I also viewed the source code of cdata_escape so that I knew what the function did: HTML encode the only argument. I compared the output before and after cdata_escape, and couldn’t see any difference. Then it’s clear that using an HTML tag for a regular expression match wasn’t feasible—change it back to Octopress’s image tag instead.

Ruby code learnt

• body.gsub!(/(?=\b).+(?=\b)/) {|m| subs[m]} from Stack Overflow.
  • /(?<=\b)/ didn’t work, since Perl and Ruby didn’t allow variable-length patterns.
  • This is for an array of Strings, not a long String.
• map and collect methods in Array class are the same.
• each and each_pair methods in Hash class are different.

An illusory success

I wrote a function, and pushed it onto GitHub.² Viewing the site, I could see that I successfully change the popup images into ordinary ones in the category Atom Feed. However, these changes were also applied to my blog posts.³ I thought that the error was due to the use of gsub!, instead of gsub.

New version of my custom filter

I’ve been tired of studying code and reading documentations. Therefore, I just adopted an ugly approach—input .gsub(...) for multiple times. Eventually, I got the desired output.