In Listing 3, lines 25 to 49 are quick to summarize: They open $RSS_ARCHIVE/latest-news, which again contains lines such as

040-FOSS|Alchemy Is A Free PDF File Converter|https://medevel.com/alchemy-file-converter/

Then lines 28 and 29 remove newlines and split each record using the pipe character as a separator, saving each column in a different variable. If the URL of the current line already exists in the %OLD_URLS hash, it means that the page was already seen, sometime in the previous two months, so the script jumps to the next news (line 30). If the URL was never seen before, lines 31 to 35 clean up the title, URL, and channel name, removing non-ASCII characters, user-tracking strings (line 34), and the numeric prefixes from channel names, which aren't needed anymore (line 35).

If the channel name has changed with respect to the previous record in $RSS_ARCHIVE/latest-news (line 37), the script saves it in $CURRENT_CHANNEL, prints it out as a level-4 HTML header (line 39), and resets the $BLOCK_COUNTER variable. This is what partitions the output in Figure 10 in different sections titled "World," "TechInnovation," and "FOSS." The <ul> and </ul> strings in line 39, and later in line 46, are just the HTML markup for the beginning and end of an unordered, bulleted list.

The single headlines are formatted as list elements and appended, just like headers and list delimiters, to the $LATEST_NEWS variable (line 43). Once all latest-news has been scanned, line 51 prints the HTML page all at once to standard output, with the list of headlines in $LATEST_NEWS in the right place.

The purpose of lines 44 to 47 is cosmetic but important, at least for me: Every seven items (line 45), the script closes the current list, inserts a blank line, and starts a new one (line 46). This simple trick greatly improves the overall readability of Figure 10 when it contains lots of news because visually separated small lists are easier to scan than just one big wall of text.

Speaking of readability, and to stress the time savings that RSS makes possible even with just a few lines of code: Note that sorting the headlines alphabetically (line 52 of Listing 2) makes them much faster to scan. In the Oscars or Super Bowl weeks, for example, there will surely be many headlines starting with those words all around the web. But if those headlines are grouped together, regardless of which website they came from, both spotting and ignoring all of them will be much quicker.