At the time of writing, script4rss is somewhat useful but quite primitive. The creation of a filter will be described here as a small tutorial. The documentation applies to the 0.4x series.

Script4rss takes a description file as input. This file describes the name, link, etc. of the feed, trhe rules to convert the HTML, and the author, license, etc. of the generated script. As of version 0.4a, script4rss only partially checks the validity if the input file

• Strings and numbers are given literally.
• Booleans can be "true", "false", "yes", "no", "1", or "0" (without the quotes).
• Regular expressions should be writte as they are used in perl, i.e. between forward slashes, modifiers can be added.
• Comments are prepended by a hash mark (#).
• Some values can extend over multiple lines. This can be done by putting the variable name and colon on all these lines.
• Hash marks and back slashes need to be escaped with a backslash.
• Additional searches for keywords may be introduced. These can be used to keep track of things like the catagory, base uri, etc.

An example

First the stuff about the generetad script is declared, like author, license, etc. A comment at the start of the script is generated with this information

Then, general information about the feed is declared.

If you know RSS, you know which elements are optional and which are required (script4rss tries to check this but I can't guarantee it works). As you can see, script4rss ingnores empty or commented values.

The update interval (feed_interval) can be in the format of XXwXXdXXhXXm to specify the number of weeks, days, hours, and minutes respectively. Other combinations are also possible, for example 1h40m if you want the feed to be updated every one and a half hour. It should be noted however, that liferea excepts at most 10,000 minutes as update interval, so 1 week is capped to 6 days, 22 hours, and 40 minutes :-P

Unfortunately, if you take a close look at the HTML code, there are some unclosed tags. We can't really on these as markers as they are maybe fixed in the feature (or indeed, in the same issue). That's why we want to filter everything through a cleanup function.

Now we get to the real stuff. You can specify multiple catagories based on different HTML uses. Unfortunately, script4rss is not smart enough (yet) to figure out how to parse the document for the catagorie an item belongs to, like it is the case here (Essay, Features, etc.). You have to give every catagory a name in between square brackets however, even if there's only one.

As you can see there are several regexes above. The first one is "search". This is a regex which is typical for the beginning of a news story. It does not need to be the beginning of a news story, it can also be an announcement (unfortunately script4rss cannot read back lines -- yet). Script4rss reads everything which should be used in an item into a single line ended whenever "match_end" is found. In this case, a closing </div>. If the news item should begin later, "match_start" should be specified. Then the final match: this is a regular expression with certain match groups (5 in tihs case). These saved groups are used for the construction of the final item,which can consist of a title, link, and description. Each of these can be specified by using the number of the match group and an optional "_prefix" and "_postfix" (as for the title), or by using backreferences in a string directly (as for the description). A combination is also possible.