|
Being a text-based writing adventure, and me being me,
I immediately set out to make Cosmic sit primarily on gopher
rather than web. At first I didn't plan to have a web presence for
it at all, but that changed as things went along. Really, my goals
and understanding of the system are ever-evolving and greatly
informed by the active users.
Regardless, things started with gopher. I knew I wanted a few
things right away:
- A single gopher structure, not individual user folders.
- Users to be able to fully control their ships and stories, but
not mess with other users' things
- An ongoing log that linked everything together.
I use motsognir [2] on gopher.black, but I thought it might be
easier to write scripts as gophermaps for cosmic if I stuck with
gophernicus. I could have made things in cgi scripts, but I like
the transparent heirarchical structure I get to this way instead. |
|
The structure of cosmic's gopher hole looks like this:
├── gophermap
├── intro.gophermap
├── listing.gophermap
├── log
│ ├── gophermap
│ └── intro.gophermap
├── Melchizedek (or any old ship)
│ ├── 001.txt
│ ├── 002.txt
│ ├── 003.txt
├── Melvin P Feltersnatch (another example ship)
│ ├── 001.txt
│ ├── 002.txt
│ └── 003.txt
├── rss.xml
└── ships
├── gophermap
├── Melchizedek (example ship page directory)
│ └── gophermap -> /var/gopher/ships/ship/gophermap
├── Melvin P Feltersnatch (another example ship page dir)
│ └── gophermap -> /var/gopher/ships/ship/gophermap
├── ship
│ └── gophermap
└── ships.gophermap
At the root level I have a gophermap that pulls in and processes
other gophermaps, does a little shell scripting, and displays the
basic content. Let's start by having a look at that code:
=intro.gophermap
1Complete Transmission Log /log
1Ships, Colonies, Outposts /ships
0RSS Feed /rss.xml
Most recent (20) log entries:
=head -n 20 listing.gophermap | sed 's|^0||' | awk -v tot="$(wc -l listing.gophermap)" '{print 0 int(tot)-NR+1 " >> " $0}'
Gophernicus allows us to source in other gophermaps by using
saying =gophermapname. In this way I can keep my intro text for
cosmic in a separate file (which becomes helpful later when we
look at web generation). Following that I have a few regular
gophermap links using gophernicus' friendly shorthand syntax (no
servername or port required). Finally I list the most recent log
entries.
In this, I grab the top 20 lines from listing.gophermap and number
them properly.
listing.gophermap is the heart of the cosmic log system. Each time
a user adds a log to the site, it gets prepended to that file.
That file in turn generates the log page, the short listings here,
and content for the individual ship pages. It's a sensitive file
because it needs to be writable by all users on the system. I've
kicked around better ways to do that, but for now it seems to
work. At least I back it up regularly.
The /log/ page works very much like the root gophermap, but with
a simpler rendering.
=intro.gophermap
=sed 's|^0||' /var/gopher/listing.gophermap | awk -v tot="$(wc -l /var/gopher/listing.gophermap)" '{print 0 int(tot)-NR+1 " >> " $0'}
Simple, right? awk is the best.
So what about the ship pages? Those are a little funkier. On the
one hand, they pretty much just sed/awk their way through the
listing.gophermap file as well, but there's also the matter of
a navigable directory.
From the list above you can see that there's a ships directory.
Lets look more closely at an example:
└── ships
├── gophermap
├── Enterprise NCC-1701
│ └── gophermap -> /var/gopher/ships/ship/gophermap
├── ship
│ └── gophermap
└── ships.gophermap
The root gophermap of the ships page does a little more hard work.
This has a few long lines that might not render well on your
screen. I apologize!:
#!/bin/sh
cat "ships.gophermap"
find "/var/gopher/" -maxdepth 1 ! -path "/var/gopher/" ! -path "/var/gopher/ships" ! -path "/var/gopher/log" -type d -print | sed 's|/var/gopher/||' | sort | while read -r ship
do
entry_num=$(grep -c "^0${ship}" "/var/gopher/listing.gophermap")
if [ "$entry_num" != "0" ]; then
printf "1%s (%s)\\t/ships/%s\\n" "$ship" "$entry_num" "$ship"
fi
done
We start by catting out the intro for the ships page from
ships.gophermap, then we loop through all the ship directories on
the system. Using each of those ships, we then check against
listing.gophermap to see how many entries they have. If they don't
have any entries, we leave them out of the listings. That was
a nice idea from one of the cosmic users to avoid cluttering up
the ship page with empty ships. If there ARE entries, we print
a link to the ship page.
The ship pages themselves are a bit of trickery. There's really
only one ship page /ships/ship/gophermap. All the rest are just
symlinks to it. But how!!?? What manner of sorcery is this?
#!/bin/sh
ship="$(/bin/pwd -L | sed 's|.*/||')"
desc="/var/gopher/${ship}/.description"
if [ -f "$desc" ]; then
cat "$desc"
printf "\\n"
fi
printf "%s - Ship Log\\n" "$ship"
tac "/var/gopher/listing.gophermap" | sed "s|^0||" | awk '{print 0 NR " >> " $0}' | grep "^.*>>\\ ${ship}"
pwd -L gets the path of the current symlink, which reveals the
ship name! Then we look to see if there's a ship description and
finally we grab all the relevant logs from listings and number
everything.
Huzzah!
That's it. That's the whole gopher structure that drives Cosmic
Voyage. Next time I'll talk about the web generation and RSS
feeds.
If this inspired you to join the system, instructions can be found
on cosmic's home page. See you in the stars! |