The Lightsabre Project
======================
Part 004 - Parsing Server Responses
New that I'm receiving data from the server, I need to make sense of
the stuff that's arriving. The gopher data is landing in a single
large buffer, and in order to display it properly I need to split it
by line, and parse each of those lines to extract types, selectors
etc.
As mentioned in the previous update, UNSTRING will parse the data for
me. It works like this:
unstring read-buffer (1:recv-bytes-read)
delimited by X"0D25"
into current-line
with pointer line-pointer.
'read-buffer' contains the data read from the server (post
ASCII-EBCDIC conversion), and 'recv-bytes-read' hold a count of bytes
received from the server. The (1:...) bit is like an array slice of
the read buffer.
Unstring works its way through read-buffer until it reaches data that
matches the 'delimited by' clause - in this case EBCDIC CRLF. Once it
reaches a delimiter, it puts the data read so far in to
'current-line', and sets 'line-pointer' to the current index of the
delimiter in the read-buffer.
At this point, we have a line ('current-line') ready for parsing. The
line parsing is also an UNSTRING, this time looking for TABs.
Wrap all this in a loop, and you're pretty much set. There's a little
bit more to it than that (tracking UNSTRING pointers, getting field
lengths etc.) but not much. Once we've got the server response parsed,
display is just a matter of looping over the parsed data.
I'm now creating permanent executables with my compiles, not the
temporary ones used previously. This is mainly because I'm now at the
stage where I need to start introducing some interactivity, so running
the thing in batch isn't an option any more. Creating executables that
remain on disk requires creating a dataset to accept them (I use
'LEE.GOPHER.LOAD') and modifying the compile JCL to use IBM procedure
IGYWCL (CL is 'compile and link'), and instructing the linker to put
it's output in my dataset.
The resulting executable can then be run in TSO. First however, I need
to tell MVS to allocate SYSOUT and SYSIN to my terminal using
'allocate'. That stuff all happens automatically in a Unix shell of
course (STDOUT, STDIN, STDERR and all that). The executable is run by
issuing the CALL command, followed by the name of the executable. When
the thing's ready for release, I'll need to do some work around
getting rid of the allocation step and making it so you can just enter
the command as eg. 'gopher', but for the time being, this is fine.
So, a copy and paste of the output as it stands is shown below:
('READY' is the TSO prompt)
-------
READY
allocate ddname(sysout) dsname(*)
READY
allocate ddname(sysin) dsname(*)
READY
call gopher.load(gopher)
gopher for mvs
hostname-value republic.circumlunar.space
hostname-length 0026
read: retcode = 0, done.
counting lines in 00002870 bytes
line count 0059
splitting data into 0059 lines
Welcome to the Mare Serenitatis Circumlunar Corporate Republic
. : /
0 .
. . / : *
/-----------------:
==================-+-+-+-+-+-+-+-+-+-+==================
++++++: /:/:/:/:/:/:/:/:/:/:/:/:/:/:/:/:/://++++++
:_______ .oo. . . . .oo._______/
:_______ oooo _______/
. | :||||/ | .
| :++++: |
* ::: |><| :::
. (((II))) *
. /::: .
. :::/
|| .
(((((oo)))))
* : / !! .
. . YY
/ : /:
/><:
!----------|UU|----------!
|__________: /__________|
. ++
.
. .
.
001 DIR About this Server
002 DIR Our Sister Colony Zaibatsu
003 FILE Signup for a Republic user account
004 DIR Recent Phlog Updates
Sundogs granted asylum here:
005 DIR aidanh010
006 DIR bacterio
007 DIR ben
008 DIR cleber
009 DIR emar
010 DIR katolaz
011 DIR kevin
012 DIR kst
013 DIR leeb
014 DIR lntl
015 DIR mijk
016 DIR mrguilt
017 DIR np89
018 DIR papa
019 DIR rond
020 DIR spring
021 DIR uwu
Please enter command (h for help)
q
done!
READY
-------
Starting to look more like a proper client now :-)
As the fancy screen handling etc. is still some way off, for the time
being I'm temporarily putting in a prompt for the user to issue
commands to the client. Not put too much thought into this yet, but
you can see above that the 'q' command seems to work pretty well.
There are numbers next to the selectors, which I've added in
anticipation of implementing, say, a 'g' command (goto), which will
accept a number parameter to follow a particular selector. So eg.
'g 017' (or something similar) would follow the link to np89's
gopherhole. You get the idea.
That's the plan, at least. Let's see how it goes.