RSS and Atom Feed Monitor

Reads any number of RSS, Atom or RDF feeds and returns their items, marking the ones that were not there on the previous run.

What it does differently

A feed that refused the request is not a feed with nothing new. That is the whole point. Most feed readers hand back an empty list when a server answers 403 or times out, and on a schedule that reads as "no news today". Here a failed feed keeps its previous snapshot untouched and the run says which feed failed and why, so an outage never turns into a false quiet spell or a flood of items reported as new the run after.

It says how sure it is that an item is new. Deciding whether an item is new means having a stable identity for it, and feeds do not reliably give one. Some publish a proper guid, some publish a URL that changes when the site does, and some publish nothing at all. Every row carries idSource naming which was used, so a row identified by guid and a row identified by its position in the document are not presented as equally trustworthy.

A cap is never invisible. Every row says how many items the feed actually held, so a capped run cannot be mistaken for a quiet one.

Input

Feed URLs, a per-feed cap, and a switch for returning only what is new. The first run on a feed returns everything with isNew empty, because nothing has been seen to compare it against.

Output

One row per item, with the feed's own title and format alongside the item's title, link, author, dates and text.

Dates are the other place feeds mislead. An item whose date cannot be believed, such as 1970 or a year in the future, is returned as null rather than as a date, because a wrong date sorts to the top of your table and reads as real.