#!/usr/bin/perl
# This is a demo script which illustrates how to write a
# special-purpose CVS client. It is a CGI script which converts the
# output from "cvs log" to XML/HTML for easy access from ECMAscript/DOM (for
# example). Note that you need a CVS with a tagged text patch for "cvs log".
# It was written by Cyclic Software, http://www.cyclic.com/, and is in
# the public domain.
$cvsroot = '/home2/cvsroot';
# Make sure to pick a file with at least one log message longer than
# one line.
$file = 'cyclic-admin/web/cvs/dev-res.html';
use FileHandle;
use IPC::Open2;
# For demo purposes, we just run the "cvs server" command as a subprocess.
# This might be sometimes useful in Real Life(TM) but it would be
# also relatively easy to use perl sockets to open a pserver/rsh/etc
# connection. Alternately, just change "cvs server" to "rsh foo cvs server".
# Or change it to "tee log | cvs server" for debugging.
$pid = open2( \*Reader, \*Writer, "cvs server" ) || die "cannot open2: $!";
Writer->autoflush(); # default here, actually
# OK, first send the request to the server. A simplified example is:
# Root /home/kingdon/zwork/cvsroot
# Argument foo/xx
# Directory foo
# /home/kingdon/zwork/cvsroot/foo
# Directory .
# /home/kingdon/zwork/cvsroot
# log
# although as you can see there are a few more details.
print Writer "Root ", $cvsroot, "\n";
print Writer "Valid-responses ok error Valid-requests Checked-in Updated Merged Removed M E MT\n";
# Don't worry about sending valid-requests, the server just needs to
# support "log" and if it doesn't, there isn't anything to be done.
print Writer "UseUnchanged\n";
print Writer "Argument ", $file, "\n";
# The protocol requires us to fully fake a working directory (at
# least to the point of including the directories down to the one
# containing the file in question). Not sure this is still true as
# of CVS 1.10.1.
# So if $file is "dir/sdir/file", then @dirs will be ("dir","sdir","file")
@dirs = split (/\//, $file);
$path = "";
foreach (@dirs) {
if ($path eq "") {
# In our example, $_ is "dir".
$path = $_;
} else {
print Writer "Directory ", $path, "\n";
print Writer $cvsroot, "/", $path, "\n";
# In our example, $_ is "sdir" and $path becomes "dir/sdir"
# And the next time, "file" and "dir/sdir/file" (which then gets
# ignored, because we don't need to send Directory for the file).
$path = $path . "/" . $_;
$file = $_;
}
}
# The "admin" request seems to require us to send "Entry". I don't
# think it matters what the contents of the Entry are; perhaps it is a
# bug that it requires it in the first place.
# Apparently we don't need it for "log".
#print Writer "Entry /", $file, "/1.1///\n";
# And the last "Directory" before the command is the top level.
print Writer "Directory .\n";
print Writer $cvsroot, "\n";
print Writer "log\n";
# OK, we've sent our command to the server. Thing to do is to
# close the writer side and get all the responses. If "cvs server"
# were nicer about buffering, then we could just leave it open, I think.
close (Writer) || die "cannot close: $!";
print "Content-type: text/html\n\n";
print "Demo of DOM and CVS\n";
print "\n";
print "\n";
print "
";
# Ready to get the responses from the server.
# For example:
# E Annotations for foo/xx
# E ***************
# M 1.3 (kingdon 06-Sep-97): hello
# ok
while () {
@words = split;
# Adding one is for the (single) space which follows $words[0].
$rest = substr ($_, length ($words[0]) + 1);
if ($words[0] eq "E") {
# Not doing anything fancy is a bit of a kludge. This _is_ a
# demo script, so I'm sure anyone would be disappointed with more :-).
print "Error message follows:\n";
print &htmlify($rest);
} elsif ($words[0] eq "M") {
print &htmlify($rest);
} elsif ($words[0] eq "MT") {
if (substr ($words[1], 0, 1) eq "+") {
&start_tag (substr ($words[1], 1));
} elsif (substr ($words[1], 0, 1) eq "-") {
&end_tag (substr ($words[1], 1));
} elsif ($words[1] eq "newline") {
print "
\n";
} else {
&start_tag ($words[1]);
# Adding one is for the (single) space which follows $words[1].
$text = substr ($rest, length ($words[1]) + 1);
# Remove the trailing newline.
chop($text);
print &htmlify ($text);
&end_tag ($words[1]);
}
} elsif ($words[0] eq "ok") {
# We could complain about any text received after this, like the
# CVS command line client. But for simplicity, we don't.
} elsif ($words[0] eq "error") {
die $rest;
} else {
# This covers Checked-in, Updated, etc., which I don't think
# the server should be sending in response to this command (not
# that cvsclient.texi tries to specify this).
warn "unrecognized response: $_";
}
}
close (Reader) || warn "cannot close: $!";
print "\n";
# Here are the XML-style versions of start_tag and end_tag.
# Probably this is cleaner than .
#sub start_tag {
# local($string) = @_;
# print "";
#}
#sub end_tag {
# local($string) = @_;
# print "\n";
#}
# Here are the versions of start_tag and end_tag.
# Hopefully I can get this working faster than the XML-style.
sub start_tag {
local($string) = @_;
print "";
}
sub end_tag {
local($string) = @_;
print "\n";
}
sub htmlify {
local($string) = @_;
# I guess we want a non-breaking space. The whole issue of filling
# paragraphs is a bit fuzzy, but we don't want to lose information
# when going to HTML/XML, so we want to make sure that spaces and
# tabs and such do get distinguished from the normal whitespace
# that separates the various bits of HTML/XML.
$string =~ s/&/&/g;
$string =~ s/ / /g;
$string =~ s/\t/ /g;
$string =~ s//>/g;
$string =~ s/\n/
\n/g;
$string;
}