Offer archived pages under more failure conditions

[da2x]

This is a request similar to issue #40.

Here are some additional failure signatures and some HTTP status codes that imply that the user would be better served with an page prompt from Archive.org than the original page:

Text matches:

• “Error establishing a database connection” (exact and complete match)
• “<br />\n<b>(.*)</b>\:(.*) on line <b>[0-9]+</b><br />” (is entire document, PHP error signature matching all pages that fail and only display a PHP error for whatever reason)

Other:

• 0 byte content length (excluding headers), and only if status code is not 204 No Content

Official HTTP status codes:

• HTTP 408 Request Timeout
• HTTP 410 Gone
• HTTP 451 Unavailable For Legal Reasons
• HTTP 500 Internal Server Error
• HTTP 502 Bad Gateway
• HTTP 503 Service Unavailable
• HTTP 504 Gateway Timeout

Apache only (frequently seen on cheap shared hosting providers):

• HTTP 509 Bandwidth Limit Exceeded

CloudFlare HTTP status code extensions (very popular proxy service provider):

• HTTP 520 Unknown Error
• HTTP 521 Web Server Is Down
• HTTP 522 Connection Timed Out
• HTTP 523 Origin Is Unreachable
• HTTP 524 A Timeout Occurred
• HTTP 525 SSL Handshake Failed
• HTTP 526 Invalid SSL Certificate

[markjohngraham]

Thank you Aeyoun,

Those are GREAT suggestions. We are thinking hard about how to make this service better and your ideas are along the lines of some of what we are considering. Stay tuned for news about more/better features in upcoming releases.

[da2x]

@markjohngraham the HTTP status code checks were so easy to implement that I just made a pull request. Rebranding the extension as “The Internet Archive’s Amazing Internet Backup Plan” or something more appropriate should be handled separately.