ciphergoth.org > Software > put-homepages

put-homepages

put-homepages is a short Perl script for automatically updating fines on a remote FTP server to match a local archive. In its current incarntion it's designed for Demon Internet users maintaining the Web pages served under Demon's "homepages" service, but it could probably be turned to some more general tasks.

In its current form, it is not suitable for use by people who want a packaged-and-ready solution; don't download it unless you hack Perl, since you'll almost certainly find yourself hacking the source to make it do what you want. However, if you do hack Perl, you might find it very useful for maintaining your Web pages just as I do, and if you submit your improvements to me we can make it more generally useful. I plan to release it under the GPL, but if you need a more liberal license let me know, I'm pretty likely to be amenable.

Since writing this, I've been informed of two packages that do similar jobs, mirror and sitecopy. Mirror has an irritating, barely-OSD "patch" license; sitecopy is GPL. I didn't know about sitecopy when I started writing this, and it may turn out to be an all-round better solution. I can now use rsync to update my website anyway, so I don't use this any more.

Unsurprisingly, documentation is weak to nonexistant; if some of the source is incomprehensible then feel free to ask. Some facts about the script in random order:

• It keeps a copy of the current state of the FTP site in a separate directory, and compares the current to the desired state before connecting to the FTP server to compose a plan of how to update the site. When it connects, it checks that its local copy of the state is sufficiently accurate before proceeding.
• It uses a parser for "ls -lR" output to read the state of directory trees; this is particularly efficient when querying the remote site, but should do a reasonable job of the local side too, and keeps everything consistent.
• It will create and delete directories and files as well as updating file contents to keep the two sides consistent. However, it doesn't understand about renaming files - it will delete the old name and upload to the new name.
• Where it finds files present remotely that are not in its record of the current state, it fetches those files, unless the file it fetched last time appears to be up-to-date. The upshot of this for Demon Internet users is that it will fetch "bwusage.txt" whenever it changes.
• It can be configured to retrieve the latest version of your homepages from a repository of the marvellously clean and straightforward revision control system PRCS before updating the remote site - when used in this way, it ignores files present in the "new" directory that PRCS disclaims.
• It expects to find all the files it's going to use to have specific, fixed names in a configuration directory: "current" for its record of the current state of the archive, "new" for the desired state, "password" for the password to use on the remote FTP server, "extras" for uploading unexpected remote files, "remote-state" for a directory listing of the remote end potentially useful for diagnosing inconsistencies, and "PRCS" for the PRCS repository.
• It actually has quite a few command-line options which even have a little documentation.

I strongly recommend using the PRCS functionality of this script even if you use PRCS for nothing else; it's very simple to learn and does all the right things for this job. Here's how I currently use this script: I've created a user "homepgs" especially to manage it, and the configuration directory is simply "/home/homepgs". My dialout scripts include the equivalent of "su -c /home/homepgs/scripts/put-homepages homepgs", so every time I dial out, the homepgs user checks out the latest version of the home pages from PRCS, and if necessary updates the remote FTP site. I have a local copy of the pages in my home directory which I can edit freely, and I check it in when it's ready for release and simply wait for the next dialout from cron. Of course, PRCS is designed to handle multiple users working on the same archive with grace.

Remember: read and understand this script before you try and use it. But if you do, please let me know!

• put-homepages

ciphergoth.org > Software > put-homepages